How to Write the Methodology Chapter of an Accounting or Finance FYP Using Bursa Malaysia Data (2026)

Most methodology chapter templates assume you distributed a questionnaire. An accounting or finance FYP usually did nothing of the kind: you pulled five years of annual reports for two hundred Bursa Malaysia companies, built a spreadsheet of firm-year observations and ran a regression. Chapter 3 then has to justify a sample you did not survey, variables you did not ask anyone about, and a model whose assumptions the examiner will test you on. This guide gives ten steps for writing that chapter, each with a sentence you can adapt, in the order accounting and finance faculties at UM, UKM, USM, UiTM, UPM and UUM expect it. Our general guide to writing the methodology chapter covers the survey-based version; this one covers the archival version only.

Step 1: State the research design as archival and quantitative, and say why

Open the chapter by naming the design in one sentence. In accounting and finance the word examiners expect is archival: the data already exist in published documents and databases, you did not create them, and the design is quantitative and explanatory. Then give the reason in one more sentence, which is almost always that the research question concerns a relationship between firm characteristics that can only be observed in reported figures across many firms and years.

Example sentence: This study adopts a quantitative archival design, using secondary data drawn from the audited annual reports of companies listed on the Main Market of Bursa Malaysia, because the relationship between board characteristics and firm performance can only be observed across firms and over time in reported figures.

Step 2: Define the population and the sampling frame from Bursa Malaysia

The population is the set of companies your question is about; the sampling frame is the list you actually drew from. For most Malaysian studies the frame is the companies listed on the Main Market or the ACE Market of Bursa Malaysia at a stated date, filtered by sector using the Bursa sector classification. State the market, the date of the list, the number of companies and the sectors you kept, then explain each exclusion in its own sentence. The exclusions examiners expect to see justified are financial institutions, which are usually dropped because their capital structure and regulatory reporting differ from other firms; companies classified under Practice Note 17, which are financially distressed and would distort performance measures; companies with a financial year change during the period; and companies with missing data for any variable in any year.

Example sentence: The sampling frame consists of the 260 companies in the consumer products and services and industrial products and services sectors of the Main Market of Bursa Malaysia as at 31 December 2024; after excluding 18 companies under Practice Note 17, 9 companies that changed their financial year end and 31 companies with incomplete data, the final sample comprises 202 companies and 1,010 firm-year observations over 2020 to 2024.

Step 3: Justify the study period, including any crisis years

Five financial years is the usual minimum for a panel study at undergraduate level and gives enough within-firm variation for fixed effects to mean something. State the exact years, why they start where they do, and how you treated abnormal years. Studies covering 2020 and 2021 must say how the pandemic was handled: a year dummy, a robustness test that drops those years, or an argument that the effect is part of what you study.

Example sentence: The study period of 2020 to 2024 begins with the first full financial year after the sample companies adopted the current reporting framework and includes year dummies to absorb the effects of the 2020 and 2021 movement control orders on firm performance.

Step 4: Name the data sources and how each variable was collected

Every variable comes from somewhere specific, and the examiner wants a source per variable rather than a paragraph that says data were collected from annual reports and databases. Malaysian accounting and finance FYPs typically draw on the audited annual reports and Corporate Governance Reports lodged on the Bursa Malaysia announcements portal, from which board size, independence, audit committee attributes and ownership figures are hand-collected; on a financial database subscribed to by the university library, such as Refinitiv Eikon, Bloomberg or S&P Capital IQ, for financial statement items and share prices; on Bank Negara Malaysia statistics for interest rates, exchange rates and the Overnight Policy Rate; and on the Department of Statistics Malaysia for GDP growth and inflation. Say which database, which fields and the download date, because databases restate figures, and describe the coding sheet for hand-collected variables.

Example sentence: Board size, the proportion of independent directors and the presence of women directors were hand-collected from the Corporate Governance Reports of each company for each year, while total assets, leverage and return on assets were extracted from Refinitiv Eikon on 15 March 2026; a random ten percent of the hand-collected observations was re-coded by a second coder with an agreement rate of 98 percent.

Step 5: Define every variable in a table with measure, source and expected sign

This table is the centre of an archival methodology chapter. One row per variable, with the variable name, its abbreviation as used in the model, the exact operational definition, the data source, the prior study from which the measure was adopted, and, for independent and control variables, the expected sign of the relationship. Use measures that published Malaysian studies have used so that your results are comparable; the Asian Journal of Accounting and Governance, published twice a year by the Faculty of Economics and Management at Universiti Kebangsaan Malaysia, and Management and Accounting Review, formerly Malaysian Accounting Review and published by the Accounting Research Institute at UiTM, are the two places to look for the local convention. Return on assets, Tobin’s Q, leverage as total debt to total assets and size as the natural logarithm of total assets are the measures examiners expect unless you argue for others; because listed companies report under the Malaysian Financial Reporting Standards issued by the Malaysian Accounting Standards Board, state any line item whose definition changed during your period.

Example sentence: Firm performance is measured by return on assets (ROA), defined as profit after tax divided by total assets at financial year end, following the measure used in prior Malaysian board-structure studies, with a positive relationship expected with board independence.

Step 6: Write the regression model as an equation and explain each term

Present the model in equation form, numbered, with subscripts for firm and year, and then define every term below it in the order it appears. For a panel study the general form is a dependent variable for firm i in year t regressed on the independent variables, the control variables, year and industry effects, and an error term. If you test a moderating effect, show the interaction term as a separate equation, and type the equation rather than pasting an image so the examiner can read the subscripts.

Example sentence: The relationship is estimated using the following panel regression model: ROA_it = β₀ + β₁BIND_it + β₂BSIZE_it + β₃WOMEN_it + β₄LEV_it + β₅SIZE_it + Year_t + Industry_i + ε_it, where BIND is the proportion of independent directors on the board, BSIZE is the number of directors, WOMEN is the proportion of women directors, LEV is total debt to total assets and SIZE is the natural logarithm of total assets.

Step 7: Explain the choice between pooled OLS, fixed effects and random effects

This is the section that separates a finance methodology from a generic one. State that the panel structure allows three estimators, explain in one sentence each what pooled OLS, the fixed effects model and the random effects model assume about unobserved firm heterogeneity, and then name the tests that will decide between them: the Breusch and Pagan Lagrange multiplier test between pooled OLS and random effects, and the Hausman test between fixed and random effects. Say that the decision will be reported in Chapter 4, not made in advance.

Example sentence: Because the data form a balanced panel of 202 firms over five years, the model is estimated by pooled ordinary least squares, fixed effects and random effects; the Breusch and Pagan Lagrange multiplier test and the Hausman test are used to select the appropriate estimator, and the results of both tests are reported before the main regression.

Step 8: State the diagnostic tests and how violations will be treated

List the assumptions you will check and what you will do if each fails, in the order they are usually run: descriptive statistics and the treatment of outliers, normally by winsorising continuous variables at the first and ninety-ninth percentiles; multicollinearity through the correlation matrix and variance inflation factors, with a stated threshold; heteroscedasticity through the Breusch and Pagan or modified Wald test, and serial correlation through the Wooldridge test, with robust standard errors clustered at the firm level as the remedy; and, where relevant, endogeneity, addressed by lagging the independent variables by one year or by an instrumental variable approach if your supervisor agrees the instrument is defensible. Our decision guide on which statistical test to use explains the difference between these diagnostic tests and the hypothesis tests that follow them.

Example sentence: All continuous variables are winsorised at the 1st and 99th percentiles to reduce the influence of extreme observations; multicollinearity is assessed using variance inflation factors with a threshold of 10; and, because the modified Wald test indicated heteroscedasticity, all models are estimated with standard errors clustered by firm.

Step 9: Describe the software and the analysis sequence

Name the software and the version, and give the analysis sequence in the order Chapter 4 will follow: descriptive statistics, correlation matrix, diagnostic tests, estimator selection, main regression, robustness tests. Stata and EViews are the packages most Malaysian finance supervisors read fluently; R is accepted where the supervisor uses it; SPSS does not estimate panel models. Our comparison of statistical software for a Malaysian FYP explains what each package does and does not do.

Example sentence: The analysis is conducted in Stata 18 in the following sequence: descriptive statistics, Pearson correlation matrix, diagnostic tests, estimator selection, the main panel regression, and three robustness tests that replace ROA with Tobin’s Q, exclude the financial years 2020 and 2021, and lag all independent variables by one year.

Step 10: Close with data reliability, ethics and limitations of the data

An archival study has no respondents, so most Malaysian faculties do not require ethics committee approval, but the chapter must still contain a short section on data integrity and research ethics: that the annual reports are audited and publicly available, that database figures were cross-checked against the annual report for a sample of observations, that the data were used only for this study, and that no confidential information was obtained. Then state the limitations of the data honestly, as design decisions rather than apologies: survivorship bias if delisted firms are excluded, the restriction to one market, and the reliance on reported rather than economic values. Before you finalise, reread the proposal you submitted at the start and make sure the sample, the period and the model in Chapter 3 are the ones the committee approved, or explain what changed.

Example sentence: All data were obtained from publicly available audited annual reports and a licensed financial database accessed through the university library; a random sample of 50 firm-year observations was traced from the database to the annual report with no material differences, and no ethics committee approval was required because the study involves no human participants.

Common mistakes that send an accounting or finance Chapter 3 back

The first is a questionnaire template with the words replaced: pilot-study and reliability sections that mean nothing for archival data. The second is a sample paragraph that never reconciles from the Bursa list to the final number. The third is a variables table without a source column or a citation column, which leaves the examiner unable to tell whether the measure is standard. The fourth is choosing fixed or random effects in Chapter 3 before running the Hausman test, or never mentioning the choice at all. When it comes to writing up the estimates, our guide to the results chapter gives the sentence template for reporting a regression coefficient with its standard error and significance.

Draft Chapter 3 while the dataset is still open

Write the chapter the week you finish cleaning the data, while the exclusions, sources and variable definitions are still in your spreadsheet notes. Tesify helps you turn those notes into a structured Chapter 3 the same day: paste the sample reconciliation, the variable list with sources, and the model, and it returns the ten sections above in the formal register your faculty expects, with the citations to prior studies formatted in APA 7th edition and kept in step with your reference list. You then check every number and every claim against your own data, because at the viva the design is yours to defend.

Generate your accounting methodology chapter with Tesify and spend the saved days on the robustness tests.